system suitability (SS) USP-NF

[cromatoloco]

La USP says to run 5 replicate injections. These injections must to be consecutive or not? I think the SS must to be run at the beginning in order to be sure that the system is OK or the system precision is OK (RSD < 2.0%). But some people says that these injections could not to be consecutive, it means that they can use std injections of the bracket (std's after samples injections) in order to arrive to RSD<2.0%. For example, if in the SS for the 5 first std injections the RSD>2.0%, they can use the other std injections after the samples injections for to arrive RSD<2.0%.

So the question is to do the SS in order to validate the HPLC system before inject the samples or only for to find the precision of the method?

[mattmullaney]

Hi Gustavo,

Perhaps the five replicate injections need not be consecutive to each other within the injection queue, but generally the HPLC system must be demonstrated to be suitable Prior To Measuring Samples (and these five injections demonstrate the repeatability of the injector at a given time of use). Other system suitability criteria such as LOQ or Resolution Solutions also are to be scheduled prior to sample analysis as well as a standard check and diluent blank injections.

As to the bracketing standards (also known as continuing calibration verification or "in-run" standards), there are multiple ways to use these...really this depends on the SOP your organization uses. In my experience, the peak areas/heights for the bracketing standards were to have values within 2.0% RSD during the course of a sample run both within their own set As Well As compared with the five initial standard injections performed in system suitability.

Yes, I agree with you, system suitability is to be assessed prior to analyzing samples, and I'd go on to say further that the bracketing standards in the analysis queue are important to have, but are not a substitute for the initial five injections of the working standard.

[HPLCaddict]

For example, if in the SS for the 5 first std injections the RSD>2.0%, they can use the other std injections after the samples injections for to arrive RSD<2.0%.

Wow, for me that's really a new way to interpret that SST criterion. Any decent auditor will kick them in the ***.
If the first 5 consecutive injections fail to reach the desired %RSD, the system is NOT SUITABLE at this moment. No room for interpretations here! So the samples were run on a system which definitely failed SST, it doesn't matter what the standards after the samples look like.
Agree with Matt, the sense of an SST is to prove the system's suitability BEFORE any samples are run. Bracketing standards can only be used to prove that the system is still behaving fine over the course of the run.

[tom jupille]

While I agree that doing the SS *before* running samples is a good idea, I believe that you *can* use the check samples for the SS. There are two major problems with that, however:

1. You don't know until the end of the run if you passed or failed SS. If you fail, then all the data obtained are deemed invalid and the results are not usable.

2. You have to decide ahead of time to do this; you can't change the rules in mid-stream. So if you failed an initial SS, you cannot just keep running check samples until you pass.

All things considered, using the SS as check samples is a bad idea.

[cromatoloco]

I am in accord with all of you. Always I thought that the SS validate the HPLC system before the sample's injections (first 5 or 6) but is true too that we can use the SS for all std injections but, is a risky decision to do that.

So, what do think about the fact that "that people" wants to eliminate one injection of the SS (considering the first 5) because of the RSD>2.0%. "That people" says that they have an internal SOP (named: treatment of outlier data) that could eliminate data when one injection area is over +/- 3 x stdev of the average data excluding the outlier data (or outlier area injection). I think is ridiculous but... we can eliminate one injection of the SS with an internal SOP? The internal SOP do not say anything about SS, is a general treatment of outlier data.

[HPLCaddict]

So, what do think about the fact that "that people" wants to eliminate one injection of the SS (considering the first 5) because of the RSD>2.0%. "That people" says that they have an internal SOP (named: treatment of outlier data) that could eliminate data when one injection area is over +/- 3 x stdev of the average data excluding the outlier data (or outlier area injection). I think is ridiculous but... we can eliminate one injection of the SS with an internal SOP? The internal SOP do not say anything about SS, is a general treatment of outlier data.

Just my 2 cents:
- If the SST says %RSD from 6 injection <2.0%, you must consider 6 injections! There's NO excuse for omitting one. The SST is a vital part of each method. If it's not fulfilled, the system is not suitable. You're not allowed to circumvent any SST parameter with a general SOP!
- That "outlier SOP" is some sort of smart thing which can go terribly wrong! If they work in a regulated environment, there MUST be an investigation of outlier data. You cannot just eliminate any data because of a general SOP! Furthermore, if they use the RSD of the data to decide wether there's an outlier or not, there's one further problem: RSD starts to make sense only if you have at least 3 values. That means, including the possible outlier, you need at least 4 values. So they must generally do 4-fold injections of each sample if they want to omit one injection...

Summarized, any capable auditor will have much fun with that outlier SOP...

[krickos]

La USP says to run 5 replicate injections. These injections must to be consecutive or not? I think the SS must to be run at the beginning in order to be sure that the system is OK or the system precision is OK (RSD < 2.0%). But some people says that these injections could not to be consecutive, it means that they can use std injections of the bracket (std's after samples injections) in order to arrive to RSD<2.0%. For example, if in the SS for the 5 first std injections the RSD>2.0%, they can use the other std injections after the samples injections for to arrive RSD<2.0%.

So the question is to do the SS in order to validate the HPLC system before inject the samples or only for to find the precision of the method?

First, USP has harmonized their requirement with Ph Eur with regard to RSD for assays.

Two, FDA has been in favour of bracketing standards/SSTs for at least ten years, USP has been slow here.

Three, have had several auditors including FDA (last one summer 2012), still stuck in the old thinking ie 5-6 injections prior to samples, realizings after some thinking or contacting home office that recommendations has changed.

Four. I would recommend doing at least 3 injections prior to samples to keep an Eye on RSD, assay sequences tends to be long (especially content of uniformity) so to get an indication if you run overnight (not unusual). Bracketing the other ones, one last in sequence.

Five. While harmonized USP is more fuzzy, Ph Eur chapter 2.2.46 is more clear and scientific based on bracketing in general.

Hope I did not miss anything.

Kind regards

[krickos]

And agree with "HPLCaddict" last post.

[gtma]

I agree with krickos. I use the approach he suggested in special circumstances. Is it acceptable to inject std 2 only 1-2x for std agreement (confirm you prepared the stds correctly) and inject Std 1 at least 5-6 times....throughout the run for system suitability? For methods that has limited solution stability and long run time, I try to limit the amount of stds injected without compromising std agreement and system precision. Any thoughts?

[cromatoloco]

So, what do think about the fact that "that people" wants to eliminate one injection of the SS (considering the first 5) because of the RSD>2.0%. "That people" says that they have an internal SOP (named: treatment of outlier data) that could eliminate data when one injection area is over +/- 3 x stdev of the average data excluding the outlier data (or outlier area injection). I think is ridiculous but... we can eliminate one injection of the SS with an internal SOP? The internal SOP do not say anything about SS, is a general treatment of outlier data.

Just my 2 cents:
- If the SST says %RSD from 6 injection <2.0%, you must consider 6 injections! There's NO excuse for omitting one. The SST is a vital part of each method. If it's not fulfilled, the system is not suitable. You're not allowed to circumvent any SST parameter with a general SOP!
- That "outlier SOP" is some sort of smart thing which can go terribly wrong! If they work in a regulated environment, there MUST be an investigation of outlier data. You cannot just eliminate any data because of a general SOP! Furthermore, if they use the RSD of the data to decide wether there's an outlier or not, there's one further problem: RSD starts to make sense only if you have at least 3 values. That means, including the possible outlier, you need at least 4 values. So they must generally do 4-fold injections of each sample if they want to omit one injection...

Summarized, any capable auditor will have much fun with that outlier SOP...

They use 2 preparations, 2 injections for each preparation, it means 4 injections. If one of the 4 is outlier they eliminate 1 and work with only 3 areas.

[cromatoloco]

While I agree that doing the SS *before* running samples is a good idea, I believe that you *can* use the check samples for the SS. There are two major problems with that, however:

1. You don't know until the end of the run if you passed or failed SS. If you fail, then all the data obtained are deemed invalid and the results are not usable.

2. You have to decide ahead of time to do this; you can't change the rules in mid-stream. So if you failed an initial SS, you cannot just keep running check samples until you pass.

All things considered, using the SS as check samples is a bad idea.

Consider this scenario:
- 5 injections std1 (SST)
- 2 injections std2 (in order to find the recovery) (SST)
- preparation 1 - sample 1 x 2 injections
- preparation 2 - sample 1 x 2 injections
- until 10 injections (first 10 injections)
- 1 injection std 1 (SST)
- 1 injection std 2 (SST)
- 10 injections (second 10 injections)
- 1 injection std 1 (problems in RSD)
- 1 injection std 2

We can say that the SST must pass RSD<=2% it means the RSD of all standards untill the end of the sequence must pass RSD<=2%. We calculate the assay with bracketing of each 10 injections but if we have problems we can use the data untill the problem is present. In our scenario we can use the data of the first 10 injections but not the second 10 injections or all injections are failed?

[Parham Chegini]

Hi everybody,
I think this kind of SST is suitable for preparation of working standard only, It waste time and material as much as possible (toxic materials for environment!). When you use a calibrated system and have a good program for P.M. and intermediate checks on your HPLC Column, this process is not necessary as this type, I am agree that check RSD for 6 Injections of STD is enough.
Please be more tender with our environment for our future race.

[HPLCaddict]

Hi everybody,
I think this kind of SST is suitable for preparation of working standard only, It waste time and material as much as possible (toxic materials for environment!). When you use a calibrated system and have a good program for P.M. and intermediate checks on your HPLC Column, this process is not necessary as this type, I am agree that check RSD for 6 Injections of STD is enough.
Please be more tender with our environment for our future race.

This is a noble and honorable attitude, but I'm afraid that environmental concerns are not really a strong argument for auditors or inspectors .
And while system calibration and regular preventive maintenance are definitely a good idea and a must in regulated environments, they are definitely no subtitute for a decent SST.

[Harish ram]

Why for precision studies only 6 injections should be given? & for system Suitability only 5 injections ? What is the logic behind the digit 5 and 6?

[cromatoloco]

It depends in your RSD. For example for the USP. If the RSD is lower than 2.0 you need 5 injections. If your RSD is greater than 2.0 you will need 6 injections as minimum. If you need better system suitability you could decrease your RSD and decrease your number of injections.

Why for precision studies only 6 injections should be given? & for system Suitability only 5 injections ? What is the logic behind the digit 5 and 6?